The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes metadata and properties from external asset catalogs. It mitigates injection risks by limiting the amount of data returned to the agent via bounded context receipts and by providing clear instructions for the agent to treat asset content as untrusted data.
[DYNAMIC_EXECUTION]: The code uses __import__ solely to access the sys module for standard error output, which is a common and benign Python practice.
[COMMAND_EXECUTION]: Local management commands perform comprehensive path validation, ensuring that all file operations are confined to the intended fixture directory and explicitly rejecting symlinks to prevent unauthorized file access.

open-asset-fixture